|Publication number||US7844415 B1|
|Application number||US 12/017,161|
|Publication date||Nov 30, 2010|
|Filing date||Jan 21, 2008|
|Priority date||Aug 20, 2007|
|Also published as||US7826999, US8032324|
|Publication number||017161, 12017161, US 7844415 B1, US 7844415B1, US-B1-7844415, US7844415 B1, US7844415B1|
|Inventors||Tyler M. Bryant, Davy J. Figaro, Andrew T. Taylor|
|Original Assignee||Pni Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (13), Non-Patent Citations (3), Referenced by (45), Classifications (12), Legal Events (2)|
|External Links: USPTO, USPTO Assignment, Espacenet|
Applicant claims priority from U.S. Provisional application Ser. No. 60/956,907 filed on Aug. 20, 2007.
The prior art for the present invention is found in U.S. Pat. No. 4,851,775, U.S. Pat. No. 5,239,264, U.S. Pat. No. 5,381,603, U.S. Pat. No. 6,243,660, U.S. Pat. No. 6,549,145, U.S. Pat. No. 7,451,549, filed on Aug. 9, 2006 and U.S. Provisional application Ser. No. 60/956,907 filed on Aug. 20, 2007. Disclosure from these patents and applications is incorporated by reference and included herein in their entirety.
1. Field of the Invention
The invention relates generally to the determination of orientation through sensing Earth's magnetic field, and more particularly to a method and apparatus to correct compass magnetic heading and/or position sensing indications to account for motion and/or magnetic distortions.
2. Description of Related Art Including Information Disclosed Under 37 CFR 1.97 and 1.98
Orientation instrumentation often uses acceleration or tilt sensors to determine the tilt angle relative to earth horizontal reference. Most of these tilt sensors are sensitive to motion acceleration, distinct from the acceleration of gravity, which can result in large orientation errors during periods of movement. The orientation error is directly seen in pitch and roll errors, as well as indirectly seen in yaw errors.
One means of compensating for motion induced orientation errors which has been previously employed is the use of the gyroscopic effect to monitor rotation rates and correct for the motion induced errors, since gyros are not generally subject to the same errors. This method has the disadvantage in practice that accurate and stable gyros tend to be large, expensive, and have relatively large power requirements. Inexpensive gyros tend to have high drift rates, limiting their usefulness to short duration motions. In either case, gyros add additional complexity to an orientation system.
Previous patents cover various methods of integrating gyros into compassing and orientation applications, such as U.S. Pat. No. 6,842,991 by Robert Levi and Robert Marshall, issued 18 Jan. 2005. Clymer Technologies disclosed U.S. 2004/0017192, now abandoned, as an orientation sensor with four magnetic sensors and a method for calculating magnetic inclination. U.S. Pat. No. 6,842,991 and U.S. 2004/0017192 are incorporated by reference herein in their entirety.
Clymer Technologies' Terrella6™ offers a heading and motion sensor capable of determining roll, pitch, yaw, translational accelerations, angular rates, and compass heading. The sealed device contains no moving parts and has interference detection to reduce ferrous metal disturbances.
The method disclosed by the instant invention makes use of an ambient magnetic field as a rotation reference during periods of motion.
Orientation instrumentation often uses acceleration or tilt sensors to determine the tilt angle relative to earth horizontal reference. Most tilt sensors are sensitive to motion acceleration distinct from the acceleration of gravity, which can result in large orientation errors during periods of movement. The instant invention discloses a method for isolating and calculating gravitational and non-gravitational components of acceleration from accelerometer measurements by using an ambient magnetic field, typically the earth's magnetic field, as a fixed rotation reference. Using magnetic field measurements, one can track changes in orientation of a device and use that information to accurately determine the direction of gravitation, even during periods when acceleration measurements include non-gravitational acceleration combined with gravitational acceleration components. In addition to accurately determining orientation, a method provides a non-gravity acceleration vector of the device. This non-gravity acceleration vector can then be used for a variety of applications including, but not limited to, inertial navigation (dead-reckoning), position determination, gesture based controllers, video game/computer interfaces, and autonomous vehicle control.
In a tilt-compensated compass, the instant invention provides a solution to motion induced orientation errors by correcting a tilt reading during periods of movement and calculating a tilt compensated heading reading, thus giving accurate orientation readings. Using a local magnetic field provides an independent rotation reference without the expense and limitations of one primary alternative, gyroscope integration.
In one embodiment, a tilt compensated compass comprises at least two and optionally, three magnetic sensors (x, y, z) and at least one and optionally, two or three accelerometer sensors (x, y, z). When the compass is not being accelerated, accelerometers provide an accurate reference of horizontal for determining a north component of the earth's local magnetic field. Upon moving a tilt compensated compass, accelerometer readings include a motion induced component in addition to an earth gravity component, resulting in orientation error. However, if a tilt compensated compass traces the path of a three dimensional, 3D, earth magnetic field vector (B), a tilt compensated compass can sense and track the direction and magnitude of the rotation which has occurred since the last measurement. Sensing a rotation recorded by B, a tilt compensated compass can use this to isolate the non-gravity accelerations (NGA) and the gravitational vector (G) from one or more accelerometer measurements (Gmeas).
Consider the case of a tilt compensated compass undergoing lateral acceleration. The orientation is not changing, so real tilt and yaw remain unchanged, but accelerometer readings cause the tilt readings to change. Because no rotation has occurred, the 3D B vector remains unchanged. Therefore one can infer that the tilt is unchanged and ignore the corrupted tilt indication.
Pitch & Roll Acceleration
In the case of a pitch & roll motion, B serves as a tilt reference which is not affected by acceleration.
Tracking the Rotation Axis
One difficulty in implementing an algorithm is to find the axis of rotation among magnetometer readings. A direct rotation, assuming a rotation axis which is orthogonal to both the first and second B field vector components will not necessarily be the correct axis, and will not account for rotation motion around the B field.
One implementation for finding the correct axis of rotation is to use a series of measurement points to track the motion of the magnetic vector. For instance, a series of three measurement points resulting from a rotation around the z-axis is shown in
Thus, in the limit of small rotations, Rd becomes Rn, the normal to a rotating plane defined by B and Ra. Once Rn is known on two adjacent rotation segments, Ra can be calculated directly as the cross product between the two vectors:
Ra[t1]=Rn[t2]ŚRn[t1] [where x≡cross product] (2)
In the case where rotations are not necessarily small, but where the rotation axis Ra is fixed, then Rd will not in general equal Rn, yet the rotation axis can still be determined by two adjacent Rd's, representing three B measurements:
The angle to rotate (gamma) can be calculated as:
gamma=a cos(dot(Rdn[t2],Rdn[t1]) (4)
where Rdn is normalized Rd:
Gamma is defined as the angle through which B[t2] needs to rotate around the axis Ra to meet B[t1]. And the corresponding rotation matrix:
Tr[t1]=rotation matrix(Ra[t1],gamma) (6)
Tr[t1] is defined as the rotation matrix for rotation of angle gamma about axis Ra.
Starting out with an initial set of magnetometer readings and a set of accelerometer readings without non-gravity acceleration distortions:
Gi=known good accelerometer readings without substantive motion distortions.
Bi=magnetometer readings taken at the same time as Gi.
Subsequent to initial good readings, a motion involving any arbitrary linear acceleration as well as a rotation around a fixed axis distorts the next series of accelerometer readings. The accelerometer measurement (Gmeas) is not used directly as G for tilt and heading calculations. Instead, an estimated G vector called Gesti, is calculated from an initial Gi and a series of Ra, gamma rotations are calculated from a series of B readings.
For a rotation on a fixed axis over 3 B readings (see
Gamma[t1]=a cos(dot(Rdn[t2],Rdn[t1])) (9)
Tri[t1]=rotation matrix(Ra[t1],Gamma[t1])=rotation matrix for rotation of B[t2] to B[t1] (10)
Update the estimated accelerometer vector based on Tri rotation:
This series of steps is repeated with each new B reading, and assuming the rotation axis Ra is approximately fixed for the most recent 3 readings, then the estimate Gesti will approximate an accurate tilt reference.
For each new B reading Gesti is updated:
Errors in the approximation may accumulate over time, so that eventually it will be necessary to reset Gesti with a new known good Gi. Over what time period errors accumulate depends on several factors, including update rate, rate of change of the rotation axis, Ra, and magnetic sensor measurement noise.
Cross-Weighted Calculation of Ra
One cause of accumulated errors involves noisy measurements combined with a small rotation delta Rd, where Rd is on a similar order as the noise, resulting in large errors in the direction of the rotation axis Ra. One solution involves tracking Ra over time to provide weighting and smoothing to the determination of the rotation axis. Adding a series of cross products Rd[t]ŚRd[t−1] from t=t1 all the way back to some depth, or time, tn, gives a weighted average of all the individual Ra's from t=t1 to t=tn−1, with depth n, where the weighting depends on the magnitudes of the individual Rd's and the sine of the angles between them. Thus larger movements (larger Rd's) have greater weighting, which is a desired result.
Ra[t1]=(Rd[t2]ŚRd[t1]+Rd[t3]ŚRd[t2]+ . . . Rd[tn]ŚRd[tn−1]) (13)
During slower motions with closely spaced measurements averaging helps smooth out a choice of axis of rotation Ra where noise may cause Ra to have a random walk component.
This method also helps improve Ra at the stop and start of motions. Consider the original unweighted method as motion comes to a stop. Rd[t1] becomes smaller than Rd[t2], and if it is comparable to the noise, then both the axis to rotate Ra and the angle gamma are somewhat random, potentially causing large errors. The new cross-weighted method helps reduce the error in Ra by weighting to previous, larger Rd values, which follow the actual Ra more closely than the random effects induced by the more recent small Rd measurements. A similar weighting to the more important larger motions happens as motion is just starting out, as well.
Additional weighting can be applied to emphasize more recent measurements:
Ra[t1]=(a1·Rd[t2]ŚRd[t1]+a2·Rd[t3]ŚRd[t2]+ . . . a(n−1)·Rd[tn]ŚRd[tn−1]) (14)
B-Ra Plane Normal Calculation of Gamma
Another source of error is the error in the rotation angle gamma which is somewhat random when Rd is comparable to the noise. One embodiment employs a simple threshold filter—if Rd is small enough, a filter sets gamma to zero to avoid unwanted rotations.
A more comprehensive embodiment calculates cross-weighted Ra rotation axis to find the angle rotated between B[t2] and B[t1] around Ra, rather than using the angle between Rd[t2] and Rd[t1]. In the absence of noise these two methods give the same result; in a noisy environment when Rd is small the Rd method is much more susceptible to noise.
First calculate the normals to the B-Ra planes:
Then find gamma:
gamma[t1]=a cos(dot(Rnn[t2],Rnn[t1])) (17)
where Rnn is the normalized version of Rn:
Summary of method using the two enhancements for finding Ra & gamma:
For a rotation on a fixed axis over 5 B readings (n=4), calculate the rotation from B[t2] to B[t1]:
gamma[t1]=a cos(dot(Rnn[t2],Rnn[t1])) (21)
Tri[t1]=rotation matrix(Ra[t1],gamma[t1])=rotation matrix for rotation of B[t2] to B[t1] (22)
Update the estimated accelerometer vector based on Tri rotation:
One situation where the earth field B does not serve as a good reference is in the case where the compass rotates around B, as shown in
For rotation motions around B, additional information is required to fully define a tilt change. However, with ideal data, such as perfect resolution, zero noise, a rotation is only a problem if it is aligned perfectly with B. Real world applications experience a “dead zone”, defined by resolution, noise, and accuracy requirements, as shown in
Selection of Gi
There are several techniques that can be employed to incorporate accelerometer measurements Gmeas into the estimated Gest. A user could select Gi by pressing a button when the device is not moving, or the device could set Gi when contact with a GPS network is lost. The processor may also use a combination of one of more constraints to determine the fitness of the incoming Gmeas signal, and using fitness to adaptively process incoming measurements to correct a best estimate Gesti. Optionally, one constraint is the magnitude of G is always equal to 1 g.
Magnitude(G)=1.0 g (24)
Additional accelerations will cause the overall acceleration magnitude to deviate from its nominal value of 1.0 g as seen by a combination of the xyz components of accelerometers:
Magnitude(Gmeas)≠1.0 g (25)
Another optional constraint is that an angle of inclination (dip-angle) and magnitudes of G and B will be a constant under rotations, this can be computed as the cross product:
Other information that can be used to determine tilt-compensation include:
Rotation of earth magnetic field must be consistent with indicated tilt. The earth field provides a solid relative reference since the last clean acceleration tilt reference. Start with a known good G reading Gi.
Another optional constraint is to mathematically analyze raw acceleration data to separate spurious acceleration information from gravity tilt information.
Another optional constraint is to require that an acceleration calculation must be consistent with the earth field rotation. Trace the rotation of Gi based on how B has rotated. Dynamically interpret the G history to remove acceleration distortions.
Another optional constraint is to apply weighting to recent accelerometer readings to provide direct reference to accelerometers.
Additionally, to enable better accuracy, measurement timing enhancements such as timing interpolation are advantageous.
One embodiment employs an algorithm to create a “reference horizon”, as determined by periodically monitoring up to three accelerometers during relatively inactive periods when accelerometer readings are not in error. During these “quiet times” a long term filter resets a reference horizon, which defines the last-known undistorted measurement of the orientation of the device; this can be seen as resetting the pitch & roll. Subsequent orientation changes are tracked exclusively through changes in the magnetic field vector direction until the next quiet reset period. Optionally, a reset period is implemented as a weighted filter with higher weight given if there is less movement detected.
Overall Algorithm Implementation
There are a number of elements to combine into an overall estimate of true G:
1) a known good G at some previous time;
2) a series of magnetometer measurements, Bmeas, and the associated series of distorted accelerometer measurements, Gmeas;
3) a rotation matrix based on B tracing back to Gi thru Gesti;
4) a dynamic interpretation of G history to remove acceleration distortions;
5) several constraints on how G may have rotated. It is possible to assemble all these components into one equation that provides an accurate estimate of G even with acceleration distortions present.
Gest[tn]=(a0·Tri0[tn]·Gesti0[tn+1]+a1·Tri1[tn]·Gesti1[tn+1]+ . . . )+(b·Gestg[tn])+(c0·G[tn]+c1·G[tn+1]+c2·G[tn+2] . . . )+(d0·Gest[tn]+d1·Gest[tn+1]+d2·Gest[tn+2] . . . ) (27)
Gest Final best estimate for the true G
Gesti Estimate of G based on rotation of B from the time of the initial “good” reading of G. Coefficients am (a0, a1 . . . ) weight Gestim based on how good each initial Gim is considered to be.
Gestg Estimate of current G based on dynamic analysis of G history to remove acceleration distortions, with consideration of the constraints of dip angle and that earth's gravity always equals 1.0 g. Coefficient b is weighted according to how good Gestg is considered relative to the other equation inputs.
cm Weighted coefficients for most recent G readings, weighted higher if there is less distortion present. This weighting keeps some level of direct reference to direct measured acceleration.
dm Weighted coefficients for most recent Gest outputs. This can be considered a recursive weighting which emphasizes past history.
In an embodiment with a two axis magnetometer and a two axis accelerometer, three dimensional vectors Bmeas and Gmeas, are obtained by using techniques taught in the cited references, or by combining the four measured quantities with the system of equations provided by at least two of the following three constraints: the magnitude of the magnetic field vector, Bmeas, is equal to the magnitude of B; the magnitude of the gravitation vector, Gmeas, is equal to 1 g; the dip angle remains constant.
In an embodiment with a two axis magnetometer and a three axis accelerometer, a three dimensional vector Bmeas is obtained by using techniques taught in the cited references, or by combining the measured quantities with the system of equations provided by the following constraints: the magnitude of the magnetic field vector, Bmeas, is equal to the magnitude of B; and the dip angle remains constant. Gmeas can be obtained from the accelerometer measurements.
In an embodiment with a three axis magnetometer and a two axis accelerometer, a three dimensional vector Gmeas is obtained by using techniques taught in the cited references, or by combining the measured quantities with the system of equations provided by the following constraints: the magnitude of the gravitation vector, Gmeas, is equal to 1 g; and the dip angle remains constant. Bmeas can be obtained from the magnetometer measurements.
In an embodiment a first sequence of data in time from two or more magnetometers, Bmeas, is acquired and a second sequence of data in time from two or more accelerometers, Gmeas, is acquired. Upon starting the system for the first time, Gest, the estimate of G will be taken to be Gmeas.
Bmeas is then, optionally, processed through a signal conditioning functional block comprising standard digital filters designed to reduce measurement noise and system noise in the raw signal.
Gmeas is, also optionally, processed through a signal conditioning functional block comprising standard digital filters designed to reduce measurement noise and system noise in the raw signal.
Bmeas, after optional conditioning, is then used to determine a rotation matrix, Rmagest, or a quaternion, or a rotation axis and an accompanying angle, as described in cited references, and earlier in this specification. Optionally, Rmagest may be determined from a least squares fitting procedure for a set of data points to be rotated into a subsequent set of data points. This rotation matrix can then be applied to Gest to obtain Gmagest.
Gmeas, after optional conditioning, may, optionally, be extrapolated forward in time to synchronize with Gmagest, gives Gaccest. The confidence of Gaccest, Gacc_conf, is determined by taking the standard deviation of magnitudes of a number of the most recent Gmeas points.
Gacc — conf=1/(1+standard deviation of magnitudes of recent Gmeas) (29)
Optionally, other techniques to determine Gacc_conf may be employed, such as comparing the recent magnitudes with 1 g, or by comparing the difference of Gmagest and Gaccest. One may also use a confidence of the magnetometer estimate.
Gmag — conf=1/(1+residual of Least Squares fit) (30)
Gest, is then calculated by combining Gmagest and Gaccest based on Gacc_conf.
Gest=Gacc — conf*Gaccest+(1−Gacc — conf)*Gmagest (31)
In an embodiment used for compassing, Gest and B are combined using tilt compensated compassing techniques to output corrected Heading, Pitch, and Roll angles.
In an embodiment Gest and Gmeas are used to compute NGA, where
The embodiment may output Gest, B, and NGA which may be used in combination with an initial position (obtained from, for example, GPS) and an initial velocity (obtained from, for example, a speedometer, or alternatively starting from rest) to determine both orientation (Heading, Pitch, and Roll) as well as velocity by integration NGA, and position by integrating velocity.
Monitoring “measured G”, Gmeas, over a period of time provides some information about levels of distortion present, deviations of magnitude and dip angle, for instance. As has been discussed, some distortions can be removed with dynamic motion correction or additional algorithms, or weighting can be applied to reduce the influence of distortions; optionally, a combination of more than one method is employed. However, at some level of distortion, either in magnitude or duration, orientation accuracy is degraded to the point that the orientation output is questionable or outside a range of utility at that point a “G distortion flag” may be used to notify a user that orientation accuracy is no longer meeting a particular threshold or range. A distortion flag may take the form of a signal or state from a means for processing to a means for communicating to a user wherein an electrical signal or state is converted to an audio or visual or tactile signal for the user's information. Alternatively a distortion flag may initiate an algorithm to compensate a heading or initiate other functions of utility to a user.
Magnetic Field Distortions
The discussion about removing or reducing distortions in G assumes that the earth's magnetic field, B, is a constant reference which is helpful in limiting the effect of G distortions. However, if B=B+Bdist, where “Bdist” is an additional magnetic field from local gradients or transients, then the assumption of a pure undistorted B field is not valid.
In the general case of a distorted B, many of the same techniques of using B to correct G distortions can be applied in a similar fashion to using G to correct B distortions. For instance, as has been stated previously, an angle of inclination (dip-angle) and magnitudes of G and B is a constant under rotation, so if G is not distorted too much then G can serve as a reference for B when B is distorted, just as B can serve as a reference for G when G is distorted.
An algorithm correcting both B and G under distortion is disclosed; the algorithm applies weighting, to either B or G depending on which measurement has less distortion effects. A Kalman type model seamlessly combines B and G in a way that utilises the strengths of each.
The instant invention discloses techniques of correcting distortions by using the rotation of one vector (B) as a reference for correcting the rotation error of another vector (G); optionally (G) as a reference for (B); optionally, combinations of (B) and (G) for alternatively correcting the other. There is another approach which gleans additional correction capability by looking at different information present in the data stream.
As distortion presents itself in the data streams of G and B, the G and B readings follow separate paths. When there are sufficient indications that one of the paths is caused by a distortion then the paths can be traced over time to give more information about removing the distortion and following the trace back to its proper place before the distortion presented itself, yet taking into account any rotations that have occurred during the distortion. Indications of a distortion include a change in the dip angle or a change in G and/or B magnitude, as well as a path analysis to determine if B or G paths are consistent.
For instance, in the case where G is invariant, but B starts to change with components that are not direct rotations around G, then clearly a distortion is present in B. In general G will not be invariant, so a more complex analysis considering the paths of both B and G is required to extricate a B distortion from the path information.
For another example, consider when G traces a path which has components outward from the sphere, as well as tangential components. The outward components represent a distortion, whereas the tangential components, ideally, represent a rotation. Having knowledge of the path traveled gives more information to separate out tangential and outward components than looking at a single set of readings.
When six sensors are sampled sequentially, timing discrepancies among the six measurements may cause errors; a dynamic motion algorithm is more effective with control of the timing. This can be accomplished by interpolating all measurements to a same point in time. In one embodiment timing interpolation is done as follows: pick a measurement as a reference time and interpolate all other measurements to match the reference time.
The instant invention discloses multiple useful constraints on determining the tilt in a dynamic motion environment:
The instant invention may be a subsystem in a more complex navigation system comprising GPS, sonar systems, laser signals, or other means known to one knowledgeable in the field. When a GPS apparatus is present the longitude and latitude data it provides may be combined with tables and/or algorithms to calculate the declination angle and inclination angle for that spot on the globe.
A tilt compensated magnetic compass or position sensing device or an instrument for calculating a vector acceleration of gravity, G, in an ambient magnetic field may comprise one or more processors, ASIC's, memory chips and other integrated circuits as required to function as a means for processing and/or as a means for calculating and/or as a means for acquiring and storing sensor signals and calculating a tilt compensated heading and/or position orientation comprising rotation axis error and/or NGA. A tilt compensated magnetic compass or position sensing device may comprise one or more memory storage means such as memory integrated circuits and/or hard disk devices as required to function as a means for acquiring and storing or retaining at least two or more sets of sensor signal data. A tilt compensated magnetic compass or position sensing device or an instrument for calculating a vector acceleration of gravity may comprise one or more communication means such visual, audio, tactile, electronic, wireless, RF, USB or others known to one knowledgeable in the art. A tilt compensated magnetic compass or position sensing device or an instrument for calculating a vector acceleration of gravity may comprise one or more means to communicate with and/or process data from a global positioning satellite, GPS, system. The term magnetometer comprises various magnetic sensors such as Magneto-Inductive (MI), Magneto-Resistive (MR), Flux-gate, SQUIDs, Hall effect, or others known to one knowledgeable in the art; the term accelerometer comprises various types of motion sensors such as MEMS accelerometers, fluidic inclinometers, or other known to one knowledgeable in the art.
In one embodiment a method for calculating the vector acceleration of gravity, G, in an ambient magnetic field comprises the steps: acquiring a first sequence of data in time, Bmeas, from two or more magnetometers and a second sequence of data in time, Gmeas, from two or more accelerometers; calculating changes in orientation from Bmeas wherein the ambient magnetic field is used as a rotation reference; and processing the calculated changes in orientation with Gmeas to calculate the vector acceleration of gravity, G. Optionally, additional steps may be added; for instance, the step: calculating an orientation Heading, Pitch and Roll based upon said first sequence of data in time, Bmeas and said vector acceleration of gravity, G. Optionally, the step: calculating non-gravitational components, NGA, of said measured acceleration vector based on said second sequence of data in time, Gmeas, and said vector acceleration of gravity, G, may be added. Optionally, the step: calculating a velocity by integrating said non-gravitational components, NGA, of said acceleration vector, G, in time, may be added. Optionally, the step: calculating a position by integrating the velocity in time, may be added. Optionally, the step: interpolating in time of said first and second sequences of data such that interpolated data correspond to first and second sequences of synchronized measurements; wherein the interpolating is done after said acquiring and before said calculating, may be added. Optionally, the steps: calculating the magnitude of said Bmeas from said first sequence of data in time, and determining the presence of magnetic distortions based upon changes in magnitude of said first sequence of data, Bmeas, may be added.
In one embodiment a method for calculating the vector acceleration of gravity, G, and a substantially undistorted magnetic field, B, in an ambient magnetic field comprises the steps: acquiring a first sequence of data in time from two or more magnetometers, Bmeas, and a second sequence of data in time from two or more accelerometers, Gmeas; first calculating changes in orientation using B as a rotation reference; processing the calculated changes in orientation with Gmeas to calculate G; second calculating changes in orientation using G as a rotation reference; and processing the calculated changes in orientation with Bmeas to calculate B. Optionally, the step: calculating magnetic distortions, Bdist, as the vector difference of said B and said Bmeas, may be added. Optionally, the steps: determining when said magnetic distortions, Bdist, exceeds a predetermined distortion flag; and generating a signal when Bdist, exceeds the distortion flag, may be added. Optionally, the step: calculating an orientation Heading, Pitch and Roll based upon B and G, may be added. Optionally, the step: calculating non-gravitational components, NGA, of the measured acceleration vector based on the second sequence of data from two or more accelerometers, Gmeas, and the vector acceleration of gravity, G, may be added. Optionally, the step: calculating a velocity by integrating the non-gravitational components, NGA, of the acceleration vector, G, in time, may be added. Optionally, the step: calculating a position by integrating the velocity in time, may be added. Optionally, the step: interpolating in time of the first and second sequences of data such that interpolated data correspond to first and second sequences of synchronized measurements; wherein the interpolating is done after the acquiring and before the first calculating.
In one embodiment an instrument for calculating a vector acceleration of gravity, G, in an ambient magnetic field comprises two or more magnetometers; two or more accelerometers; means for calculating; means for acquiring and storing a first sequence of data in time, Bmeas from the two or more magnetometers, and a second sequence of data in time, Gmeas from the two or more accelerometers; and an algorithm comprising signal conditioning, constraint setting and matrix calculation; such that the means for calculating is operative to calculate the vector acceleration of gravity, G. In some embodiments the instrument for calculating a vector acceleration of gravity the means for calculating is operative to calculate an orientation Heading, Pitch and Roll using the algorithm, the vector acceleration of gravity, G, and the Bmeas.
In some embodiments of an instrument for calculating a vector acceleration of gravity, an algorithm comprising signal conditioning, constraint setting and matrix calculation comprises appropriate adaptations of Equations (1) through (34) in exemplary sequences as shown in
a) the magnitude of G is equal to 1 g;
b) an angle of inclination (dip-angle) and magnitudes of G and B is a constant under rotations;
c) setting G equal to a known good reading Gi;
d) mathematically analyze raw acceleration data to separate spurious acceleration information from gravity tilt information;
e) require that an acceleration calculation be consistent with the earth field rotation;
f) apply weighting to recent accelerometer readings to provide direct reference to accelerometers;
g) assume how G may have rotated;
h) dynamic analysis of G history to remove acceleration distortions, with consideration of the constraints of dip angle and that earth's gravity always equals 1.0 g;
i) the magnitude of the magnetic field vector, Bmeas, is equal to the magnitude of B;
j) the magnitude of the gravitation vector, Gmeas, is equal to 1 g; and
h) the dip angle remains constant across some period of time.
The foregoing described embodiments of the invention are provided as illustrations and descriptions. They are not intended to limit the invention to a precise form as described. In particular, it is contemplated that functional implementation of invention described herein may be implemented equivalently in hardware and/or various combinations of hardware and software and/or other available functional components or building blocks. Other variations and embodiments are possible in light of above teachings to one knowledgeable in the art, and it is thus intended that the scope of invention not be limited by this Detailed Description, but rather by Claims following.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US6285927||May 26, 1999||Sep 4, 2001||Hughes Electronics Corporation||Spacecraft attitude determination system and method|
|US6289268||Mar 2, 2000||Sep 11, 2001||Hughes Electronics Corp.||Attitude determination system and method|
|US6335617||May 5, 1997||Jan 1, 2002||Biosense, Inc.||Method and apparatus for calibrating a magnetic field generator|
|US6725173 *||Sep 4, 2001||Apr 20, 2004||American Gnc Corporation||Digital signal processing method and system thereof for precision orientation measurements|
|US6842991||Jul 11, 2003||Jan 18, 2005||Robert W. Levi||Gyro aided magnetic compass|
|US6931323||Aug 5, 2003||Aug 16, 2005||Samsung Electric Co., Ltd.||Apparatus and method of compensating for an attitude error of an earth magnetic sensor|
|US7155837||Feb 7, 2005||Jan 2, 2007||Samsung Electronics Co., Ltd.||Geomagnetic sensor for detecting dip angle and method thereof|
|US7295943 *||Dec 6, 2004||Nov 13, 2007||Samsung Electronics Co., Ltd.||Geomagnetic sensor for calibrating azimuth by compensating for an effect of tilting and method thereof|
|US7451549 *||Aug 9, 2006||Nov 18, 2008||Pni Corporation||Automatic calibration of a three-axis magnetic compass|
|US20020103610 *||Oct 30, 2001||Aug 1, 2002||Government Of The United States||Method and apparatus for motion tracking of an articulated rigid body|
|US20020171427 *||Apr 24, 2001||Nov 21, 2002||The United States Of America Represented By The Secretary Of The Navy||Magnetic anomaly sensing system and methods for maneuverable sensing platforms|
|US20030158699 *||Dec 8, 1999||Aug 21, 2003||Christopher P. Townsend||Orientation sensor|
|US20040017192||Apr 15, 2003||Jan 29, 2004||Clymer Technologies, Llc||3-axis magnetic angular orientation sensor|
|1||Caruso, Michael J.; "Applications of Magnetic Sensors for Low Cost Compass Systems"; Honeywell, SSEC; www.ssec.honeywell.com/magnetic/datasheets/lowcost.pdf [Apr. 26, 2010], 8 pages.|
|2||Cho, Seong Yun; "A Calibration Technique for a Two-axis Magnetic Compass in Telematic Devices"; ETRI Journal, vol. 27, Nbr. 3; Jun. 2005, pp. 280-288.|
|3||Cho, Seong Yun; "Tilt compensation algorithm for 2-axis magnetic compass"; Electronics Letters, Oct. 30, 2003, vol. 39, No. 22, 2 pages.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7970519 *||Sep 27, 2006||Jun 28, 2011||Caterpillar Trimble Control Technologies Llc||Control for an earth moving system while performing turns|
|US8131494 *||Dec 4, 2008||Mar 6, 2012||Baker Hughes Incorporated||Rotatable orientation independent gravity sensor and methods for correcting systematic errors|
|US8200246||Jun 12, 2009||Jun 12, 2012||Microsoft Corporation||Data synchronization for devices supporting direction-based services|
|US8223121 *||Dec 18, 2008||Jul 17, 2012||Sensor Platforms, Inc.||Host system and method for determining an attitude of a device undergoing dynamic acceleration|
|US8326561||May 20, 2008||Dec 4, 2012||Airmar Technology Corporation||Dynamic motion control|
|US8467991 *||May 8, 2009||Jun 18, 2013||Microsoft Corporation||Data services based on gesture and location information of device|
|US8515707||May 6, 2009||Aug 20, 2013||Sensor Platforms, Inc.||System and method for determining an attitude of a device undergoing dynamic acceleration using a Kalman filter|
|US8576169 *||Dec 18, 2008||Nov 5, 2013||Sensor Platforms, Inc.||System and method for determining an attitude of a device undergoing dynamic acceleration|
|US8587519||Jan 4, 2011||Nov 19, 2013||Sensor Platforms, Inc.||Rolling gesture detection using a multi-dimensional pointing device|
|US8615257||May 31, 2012||Dec 24, 2013||Microsoft Corporation||Data synchronization for devices supporting direction-based services|
|US8683707||Mar 28, 2012||Apr 1, 2014||Mike Alexander Horton||Magnetically modulated location system|
|US8700301||Jan 29, 2009||Apr 15, 2014||Microsoft Corporation||Mobile computing devices, architecture and user interfaces based on dynamic direction information|
|US8700302||Aug 6, 2009||Apr 15, 2014||Microsoft Corporation||Mobile computing devices, architecture and user interfaces based on dynamic direction information|
|US8868368 *||Sep 13, 2010||Oct 21, 2014||Performance Designed Products Llc||Motion smoothing in 3-D position sensing apparatus|
|US8868374||Jun 3, 2013||Oct 21, 2014||Microsoft Corporation||Data services based on gesture and location information of device|
|US8907893||Oct 29, 2013||Dec 9, 2014||Sensor Platforms, Inc.||Rolling gesture detection using an electronic device|
|US8957909||Jun 21, 2011||Feb 17, 2015||Sensor Platforms, Inc.||System and method for compensating for drift in a display of a user interface state|
|US9152249||Oct 31, 2013||Oct 6, 2015||Sensor Platforms, Inc.||System and method for determining an attitude of a device undergoing dynamic acceleration|
|US9200901||Jun 2, 2009||Dec 1, 2015||Microsoft Technology Licensing, Llc||Predictive services for devices supporting dynamic direction information|
|US9228842||Mar 22, 2013||Jan 5, 2016||Sensor Platforms, Inc.||System and method for determining a uniform external magnetic field|
|US9257054||Apr 13, 2012||Feb 9, 2016||Adidas Ag||Sport ball athletic activity monitoring methods and systems|
|US9316513||Jan 8, 2013||Apr 19, 2016||Sensor Platforms, Inc.||System and method for calibrating sensors for different operating environments|
|US9347968 *||Jun 29, 2012||May 24, 2016||Nikon Corporation||Electronic device and input method|
|US9459276||Jan 4, 2013||Oct 4, 2016||Sensor Platforms, Inc.||System and method for device self-calibration|
|US9500464||Mar 12, 2013||Nov 22, 2016||Adidas Ag||Methods of determining performance information for individuals and sports objects|
|US9504414||Apr 13, 2012||Nov 29, 2016||Adidas Ag||Wearable athletic activity monitoring methods and systems|
|US9661468||Oct 2, 2014||May 23, 2017||Microsoft Technology Licensing, Llc||System and method for converting gestures into digital graffiti|
|US20080073089 *||Sep 27, 2006||Mar 27, 2008||Francisco Green||Control and method of control for an earth moving system|
|US20090292495 *||May 20, 2008||Nov 26, 2009||Airmar Technology Corporation||Dynamic motion control|
|US20090318168 *||Jun 12, 2009||Dec 24, 2009||Microsoft Corporation||Data synchronization for devices supporting direction-based services|
|US20090319175||Jan 29, 2009||Dec 24, 2009||Microsoft Corporation||Mobile computing devices, architecture and user interfaces based on dynamic direction information|
|US20090319178 *||Jun 12, 2009||Dec 24, 2009||Microsoft Corporation||Overlay of information associated with points of interest of direction based data services|
|US20100095773 *||Dec 18, 2008||Apr 22, 2010||Shaw Kevin A||Host System and Method for Determining an Attitude of a Device Undergoing Dynamic Acceleration|
|US20100097316 *||Dec 18, 2008||Apr 22, 2010||Shaw Kevin A||System and Method for Determining an Attitude of a Device Undergoing Dynamic Acceleration|
|US20100145620 *||Dec 4, 2008||Jun 10, 2010||Baker Hughes Incorporated||Rotatable orientation independent gravity sensor and methods for correcting systematic errors|
|US20100174506 *||May 6, 2009||Jul 8, 2010||Joseph Benjamin E||System and Method for Determining an Attitude of a Device Undergoing Dynamic Acceleration Using a Kalman Filter|
|US20110071785 *||Sep 13, 2010||Mar 24, 2011||Thomas Peter Heath||Motion smoothing in 3-d position sensing apparatus|
|US20110163947 *||Jan 4, 2011||Jul 7, 2011||Shaw Kevin A||Rolling Gesture Detection Using a Multi-Dimensional Pointing Device|
|US20120136604 *||Jul 26, 2011||May 31, 2012||Industrial Technology Research Institute||Method and apparatus for 3d attitude estimation|
|US20120185204 *||Jul 26, 2010||Jul 19, 2012||Commissariat A L'energie Atomique Et Aux Energies Alternatives||Method for estimating the direction of a moving solid|
|US20140137630 *||Jun 29, 2012||May 22, 2014||Nikon Corporation||Electronic device and input method|
|US20140357305 *||May 30, 2013||Dec 4, 2014||Indooratlas Oy||Orientation of mobile device measuring earth's magnetic field indoors|
|US20150168953 *||Dec 15, 2014||Jun 18, 2015||Itrack Llc||Autonomous self-leveling vehicle|
|US20150241390 *||Feb 22, 2015||Aug 27, 2015||PNI Sensor Corporation||Automatically updating hard iron and soft iron coefficients of a magnetic sensor|
|US20150260543 *||Mar 13, 2014||Sep 17, 2015||Indooratlas Oy||Background calibration|
|U.S. Classification||702/151, 702/141, 73/514.16, 702/150, 73/488, 702/153|
|International Classification||G06F17/16, G01P15/105|
|Cooperative Classification||G01C17/02, G01C25/005|
|European Classification||G01C25/00A, G01C17/02|
|Jan 21, 2008||AS||Assignment|
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:BRYANT, TYLER M.;FIGARGO, DAVY J.;TAYLOR, ANDREW T.;REEL/FRAME:020392/0230
Owner name: PNI CORPORATION, CALIFORNIA
Effective date: 20080121
|Apr 3, 2014||FPAY||Fee payment|
Year of fee payment: 4